Development of systematic and combinatorial approaches for the metabolic engineering of microorganisms

Abstract

Explorations and optimizations through the genomic space are a daunting undertaking given the complexity and size of the possible search space. To approach this problem, systematic and combinatorial approaches were employed for the engineering of cellular phenotype in Escherichia coli. Initially, a computational method based on global cellular stoichiometry was employed to identify single and multiple gene knockout targets for lycopene production in E. coli. These targets led to substantial increases in lycopene production, but were limited in scope due to the nature of these models. Therefore, these approaches and targets were complemented with combinatorial searches to identify unknown and regulatory targets. When combined, these searches led to further increases of lycopene production and allowed for the visualization of the resulting metabolic landscape. A more exhaustive search was conducted in the background of eight genotypes which resulted in the formulation of the gene knockout search network. This network enables the investigation into how phenotype optimization is biased by search strategy.

(cont.) Collectively, these results demonstrated that despite the complexity and nonlinearity of genotype-phenotype spaces, most of the significant phenotypes were controlled and regulated by a small subset of key "gateway" nodes. Often, the mutations and genotypes incurred in altering global cellular phenotypes are not necessarily additive and can be quite non-linear. Effective probing of a metabolic landscape requires not only gene deletions, but also the varying (or tuning) of expression level for a gene of interest. Through promoter engineering, a library of promoters of varying strength were obtained through mutagenesis of a constitutive promoter. A multi-faceted characterization of the library, especially at the single-cell level to ensure homogeneity, permitted quantitative assessment correlating the effect of gene expression levels to improved growth and product formation phenotypes in E. coli. Integration of these promoters into the chromosome can allow for a quantitative, accurate assessment and tuning of genetic control. Collectively, quantitative phenotype-genotype analysis illustrated that optimal gene expression levels are variable and dependent on the genetic background of the strain.

(cont.) As a result, tools such as promoter engineering, which allow for a wide range of expression levels, constitutes an integral platform for functional genomics, synthetic biology, and metabolic engineering endeavors. Finally, multiple genetic modifications are necessary to unlock latent cellular potential. However, the capacity to make these meaningful modifications has remained an elusive task for cellular and metabolic engineering. The tool of global Transcription Machinery Engineering (gTME) allows one to explore a vastly unexplored, expanded search space in a high throughput manner by evaluating multiple, simultaneous gene alterations in order to improve complex cellular phenotypes. Through the alteration of key proteins involved in global transcription, cells may be reprogrammed for phenotypes of interest. Results in phenotype optimization using gTME outperformed traditional approaches to these problems, exceeding, in a matter of weeks, benchmarks achieved through decades of research. Through gTME, it is now possible to unlock complex phenotypes regulated by multiple genes which would be very unlikely to reach by the relatively inefficient, iterative gene-by-gene search strategies.

(cont.) The concept of gTME is generic and provides access points for diverse transciptome modifications broadly impacting phenotypes of higher organisms too, as further studies with yeast amply demonstrate. On the basis of these studies, combinatorial methods are generally more powerful in obtaining a given cellular objective than systematic methods due to their ability to make broader perturbations. However, properly designed search strategies which make use of both systematic and combinatorial approaches may be the best route for optimizing phenotypes.